FRC Blog - It Wasn’t Against the Rules

[Andrew Schreiber]

Originally Posted by Racer26

Well, the system already knows how many auto balls must go through.

It would need to deal with the ball not really being scored scenarios somehow though.

Ball 1 enters Line break sensor is now tripped while that ball is passing through.

Ball 2 enters. Line break sensor is still tripped from Ball 1.

Ball 1 completes traveling through sensor.

Ball 2 completes traveling through sensor.


How many balls were detected? (hint, it's less than 2)

[gburlison]

Why are they using touch screens when you could assign a ref to watch a ball and give him a hand held trigger with a button for assists, truss/catch, high, and low goal. She would never have to take her eyes of the field.

[BigJ]

Originally Posted by gburlison

Why are they using touch screens when you could assign a ref to watch a ball and give him a hand held trigger with a button for assists, truss/catch, high, and low goal. She would never have to take her eyes of the field.

refs would have to calculate assists in their head (more complex than recording zone/robot possession pairs, which would require 9 buttons)

[Alan Anderson]

Originally Posted by Racer26

Well, the system already knows how many auto balls must go through.

1) It only knows how many if someone tells it for each match.
2) Optical sensing of goal-traversing balls has been tried before (2006). It wasn't reliable. Even detecting balls that were guaranteed to come through a sensor block one at a time was problematic (2010).

[Travis Hoffman]

Originally Posted by Jared Russell

This seems to be your primary thesis. How, specifically, do you think it should change? Your very next sentence asserts that adding more volunteers does not solve anything, so you must have some idea in mind for what these fundamental changes should be.

Also, my understanding of the system is that volunteers are assigned to specific roles by local leadership (Regional Directors and the Regional Planning Committee), not by Manchester. Unless your proposal for fixing the perceived volunteer issue is to centralize all detailed Regional-level planning in NH, I am not sure how affixing your gaze towards Manchester will help in this specific instance.

This year might be a lost cause - if FIRST cannot quickly design and deploy hardware/software fixes to the system that make it easier for humans to input scoring/foul information - fixes that minimize any one individual's distraction from on-field activities - then no number of additional volunteers is going to help. The bottleneck is at the human->FMS point of interaction. I WOULD CERTAINLY HOPE THEY ARE TRYING TO FIX IT, however. Alas, the community hasn't really been updated on whether or not that is actively being pursued. I'd consider this to be of a "DEFCON" level similar to or even greater than that of the 2012 Einstein fallout, since many, many, MANY more teams are actively being affected. We praised FIRST for their openness in that situation - we need it even more right now.

Regardless of continuing FMS concerns, my contention that many referee volunteers out there are either ill-prepared or ill-suited for the role remains, independent of FMS issues. I have seen more direct video evidence of referees staring at blatant infractions (i.e. not distracted by HMI data entry at the time) than I care to recount this year.

So how to fix both in future seasons? Seems Manchester can easily invest some resources to GREATLY address these problems, with the expenditure of a little more money and acquisition of more employee manpower.

Better referee training. Earlier referee training. Visual referee training. A focus, once and for all, in ensuring CONSISTENCY in application of the rules - which has been a long standing complaint of countless teams and mentors for as long as I've been involved in this program. FIRST can spend more money to hire quality control staff to develop better, more visual training materials - referee training and any other training of key volunteers - LRI's, etc.

FIRST can also, critically, spend more money to hire more field development staff to design and build and VET better field control systems. Many hands make light work, and many brains can help identify the human interaction bottlenecks experienced with this year's FMS and ensure that they are properly eliminated.

I have asked a veteran referee - a degreed engineer - his thoughts about the overall training process. He has been thoroughly underwhelmed by the available training materials and methods over the years. I encouraged him to email FIRST HQ and explain this in detail to them, for anyone who sits back and simply accepts things the way they are is not doing anything to affect necessary change. Here is a brief summary of his feedback shared with me:

• Text-based referee training materials are insufficient to properly prepare referees for their critical role. If veteran referees are not enthused/confused by the material, that does not bode well for referees who've never done this sort of thing before.
• Many questions on the referee qualification exam are nebulous and/or confusing and do not lend themselves to ensuring command of the rules and manual.
• The lack of any visual/video support materials highlighting examples of what is and what is not an infraction is a glaring omission. The development of such things would help eliminate individual interpretation and spread consistent interpretations throughout the entire referee community. In addition, such things could be published publicly TO SHOW TEAMS A BASELINE FOR WHAT IS AND IS NOT ACCEPTABLE GAMEPLAY BEHAVIOR in regards to rules that are most open for interpretation. Show teams what to expect, and they are more likely to CONFORM TO THAT EXPECTATION, minimizing the freaking wide gamut of CRAP GAMEPLAY referees have to sift through during matches. Note the capitalization and embellished emphasis was my own and not his. Go figure.
• Video can also help a referee better understand how to interact with and operate the field control HMI's. Perhaps much more quickly and effectively than a text/pictorial based manual.
• Interaction with head referees in any pre-event Q&A sessions is often limited and insufficient to truly prepare one for the role, especially if prior training has left one with a lot of doubt as to how they should be calling key infractions.
• Pre-event conference calls with head referees are not mandatory - they are at the discretion of the head ref.

So there you go. The upfront expense to add additional human resources toward improving referee preparation and optimizing the field system to minimize referee distraction from gameplay. Better communication to teams as to what is and what isn't an infraction using video-based examples BEFORE the competitions get under way. A top-down approach to improved communication to help once and for all generate some consistency in gameplay oversight and help cement FRC as the pinnacle of PROFESSIONAL robotics competitions.

Here would be one obvious example of what I am talking about. Recognizing the FRC game scales such that the rules/penalties/manual are more complex than this competition, that simply means FIRST needs to hire sufficient staff to produce similar content for a more complex game challenge. http://m.youtube.com/playlist?list=P...u-E8Bbcg8ifQpa

[Racer26]

Originally Posted by Alan Anderson

1) It only knows how many if someone tells it for each match.
2) Optical sensing of goal-traversing balls has been tried before (2006). It wasn't reliable. Even detecting balls that were guaranteed to come through a sensor block one at a time was problematic (2010).

My understanding is that the prematch ref screens require the refs to indicate how many alliance BALLs are on the field.

To your second point and the other poster above?

I used to do volunteer work in racecar timing applications. We used a single photo beam and were able to detect 2 cars crossing the beam together, so long as they did not do so side by side by measuring the length of the occlusion. If it was longer than the longest car in the field, you knew it was 2 cars.

In the case of the 2014 FRC goals, the low are easy to deal with, as only one ball can cross the plane of the alliance wall at a time there. In the high goal, if you orient an array of sensors perpendicular to the floor, spaced across the width of the goal such that a ball will always break exactly two of the sensors, it becomes simple to distinguish multiple balls crossing the opening.

Just because we have been unsuccessful in FRC applications before does not mean it isn't possible to do.